Problems for Positivism
Quite a few objections have been raised against the views of the Logical Positivists. I will not attempt to document them all here. But I will introduce some of the more significant problems. Several of the problems with Positivism gave rise to new developments in the philosophy of science. We will introduce problems for positivism in reference to the main elements of positivism discussed the Logical Positivism note. Again, these are the verificationist theory of meaning as a solution to the demarkation problem, the positivist’s account of scientific theories, their account of laws and their account of scientific explanation.
Problems for the verificationist theory of meaning:
The verificationist theory of meaning holds that the significance of a claim is given in the possible observable conditions that would lead us to accept the claim as true or reject it as false. One problem for the verificationist theory of meaning is that it does not appear to apply to it self. The verificationist theory of meaning makes a positive claim about what it is for a claim to be meaningful. But we apparently cannot specify any possible observable conditions that would verify or refute that claim.
The verificationist theory of meaning appeals to a distinction between observational and theoretical vocabulary. Observational vocabulary can be defined by ostension (pointing at and demonstrating the observable entity you want your term to refer to). Theoretical vocabulary is to be defined in terms of correspondence rules. Correspondence rules are explicit definitions of theoretical terms, ultimately in terms of observational vocabulary. One problem for the verificationist theory of meaning is spelling out this distinction between observational and theoretical vocabulary in a manner that makes the difference clear.
First lets consider what sorts of entities are to be considered observable. Are they things, properties and events? Or are they our subjective experiences of things and events? If what is observable is just our subjective experiences of things, then there is no common shared object of various people’s observation. This leads towards 'solipsism' - the view that nothing exists but one's self and one's own experiences. So, it looks like we are better off taking the objects of observation to be objectively existing entities like properties, things or events.
Perhaps we don't observe properties themselves. I can see red things, but I can't see the property of redness. Here we should introduce the type/token distinction. Types are universals, properties or relations that can be multiply instantiated. Tokens of a universal are the individual instances of that universal. I can observe particular instances of the property of being red. So, property instances, things and events look like good candidates for the objects of observation. But notice that unlike our subjective experiences, our observation of these things is not direct, but always mediated by our senses. This will be important when we consider the theory ladeness of observation. First we will consider some early attempts to draw a distinction between observable and theoretical entities and how these attempts fail. We will look at two such attempts
1) Rudolph Carnap proposed an ease of application criterion for observable vocabulary. The idea is just that a term should count as observable if it is easily applied, and theoretical if we need to employ some more involved test to determine whether or not it applies. But there are lots of straightforward counterexamples to this proposal. Consider any of many diseases. The disease itself may involve unobservable processes at the cellular or biochemical level. It should then be considered theoretical. But long before we know the nature of a disease, doctors may be able to diagnose it with ease. 'Chicken pox', for instance, should count as theoretical vocabulary. But a doctor familiar with the disease can easily and accurately apply the term.
2) Another proposal for distinguishing observational vocabulary from theoretical vocabulary was the instrumentation criterion. On this view, we count a term as observational if it refers to some entity we can observe without instruments like microscopes or amplifiers. However, some of us require instruments to see things that other of us can see without instruments. Suppose a farsighted field biologist notices that the leaves of certain plants appear to be getting eaten but can't observe any animal consuming them. Using a powerful magnifying glass, he can observe tiny bugs consuming the leaves. On the instrumentation criterion, the biologist would apparently have to deny that the bugs are observable entities. But, his assistant with perfect vision can see them with her bare eyes. So are the bugs observable or not. We might try to appeal to the observational capacities of "normal observers". But who falls into this category?
The Theory Ladeness of Observation
The difficulty with the instrumentation criterion stems from the fact that different people are sensitive to the world to differing degrees in differing aspects. There does not appear to be any one standard we can appeal to here. We might find such differences are much less marked among peers in a scientific community. If so, appealing to this agreement that is unique to the members of this group, would be of little aid to the Positivist in proposing a general distinction between the observable and the non-observable. Rather agreement about the observable among a group of scientists that share an area of specialization suggests the theory ladenness of observation - the notion that observation claims are always understood in the context of theoretical background beliefs. Consider recognizing an arrowhead, recognizing an electron's cloud chamber track, recognizing weather conditions, or hearing that your ignition is properly tuned. It appears that observation by scientists involves background theory that is shared by members of a scientific community but not so widely shared among people in general. This points away from the positivist's verificationism and towards a holist theory of meaning. (Exercise: Would close agreement about observational claims among members of a scientific community also suggest that their branch of science appeals to individuals who are sensitive to the world to certain degrees in certain respects? If so, what are the implications of this?)
Theories
The way to test the Positivist's view of theories is to look at lots of examples of theories that have been produced by science and consider whether or not they fit the Positivist's account of theories as formal languages. Some actual theories, like Classical physics, electrostatics, and thermodynamics seem to fit the view pretty well. Theories in the "soft sciences," like Marxist theory in political science and Freudian theory in psychology are not so easy to formalize. It comes as no surprise then, that practitioners of the social sciences lead the charge in rejecting the positivist’s view of theories.
Laws and non-material conditionals
According to Logical Positivism, laws are reports of regularities that make no appeal to specific places times or things. Laws can be expressed in terms of generalized conditionals of the form "For any x, if it is P then it is Q". On this analysis, both of the following claims count as laws:
a) For any x, if it is a solid sphere of gold, then it has a diameter of less than one mile.
b) For any x, if it is a solid sphere of uranium 235, then it has a diameter of less than one mile.
But now we have a puzzle. (a) may be a true generalization, but it seems to be an accidental generalization. Intuitively, a solid sphere of gold with a diameter of over one mile would not be a violation of the laws. Rather it is just an accident of history that none exists. Not so for (b). Given the unstable nature of heavy atoms like uranium 235, a solid sphere of uranium 235 of a mile in diameter is impossible (this much uranium 235 would be well over critical mass and a spontaneous nuclear fission reaction would prevent the existence of such a sphere). What (a) and (b) show us is that some claims fitting the positivist's analysis of law are genuinely law-like while others are not. The following discussion of conditionals will reveal a bit more about the distinguishing characteristics of law-like conditionals.
Conditionals are ‘if. , then . .' claims (or any claim that can be faithfully rephrased as an 'if . , then . .' claim). Conditionals are two part claims. The part following 'if' is called the antecedent. The part following 'then' is called the consequent.
Material conditionals:
If it rains tomorrow then I will carry an umbrella.
If it’s a mammal, then it has blood.
The logically important feature of material conditionals is that their truth is fully determined by the truth or falsity their antecedents and consequents. The following table shows the truth-value of a material conditional as a function of the truth-values of its antecedent and consequent.
Truth table for material conditionals
|
P |
Q |
If P, then Q |
|
T |
T |
T |
|
T |
F |
F |
|
F |
T |
T |
|
F |
F |
T |
Material conditionals are true if either their antecedent is false or their consequent is true. The only circumstances where a material conditional is judged false is where its antecedent is true and its consequent is false. For instance, the conditional 'if it rains tomorrow, then I will carry an umbrella' is false if and only if it does rain tomorrow and I fail to carry an umbrella. The special feature of material conditionals is that they are truth functional. This just means that the truth or falsity of a material conditional is entirely determined by the truth or falsity of its antecedent and its consequent. But the truth or falsity of many conditionals is not fully determined by the truth or falsity of their antecedents and consequents. These are Non-material conditionals.
Non-material conditionals
Non-material conditionals include those stated in the subjunctive mood (if x were P, then x would be Q) or as counterfactuals (if x had been P, then x would have been Q). For instance:
If this copper rod were heated, it would expand.
If John were an amoeba, then he wouldn't have a brain.
Non-material conditionals are not truth functional. That is, the truth or falsity of a non-material conditional is not clearly definable as a function of the truth or falsity of its antecedent and consequent. To see this, note that according to the truth table for material conditionals, a material conditional is true if it has a false antecedent (regardless of the truth value of its consequent). But consider the following two non-material conditionals:
A) If copper rod x is heated, then it would expand.
B) If copper rod x is heated, then it would shrink.
Suppose rod x is made of copper and is never heated. In this case, if we try to interpret A and B as material conditionals, both come out true since both have false antecedents. But given that rod x is made of copper, only A is consistent with the laws of nature. B is inconsistent with the laws of nature.
But the mere truth or falsity of antecedents and consequences does not fully determine the truth-values of such conditionals. Even if rod x is never heated, A' is true and B' is false. What makes A' true and B' false is not the truth or falsity of their antecedents and consequents, but the laws of nature. Because non-material conditionals are not truth functional, we have no means of faithfully representing them in terms of formal systems like first order, quantified logic.
According to the Positivist’s view of laws, the logical form of a law of nature is given as that of a universally quantified material conditional as follows:
("x) (Fx É Gx)
In English we would read this as 'For any x, if it is F then it is G'. So the law that specifies the speed of light as 186,000 miles per second would be rendered as 'For any x, if it is light, then it moves at 186,000 miles per second'. Universally generalized conditionals like this are true or false purely in virtue of the way the world is as a matter of fact. If all of the things that are rays of light move at the specified speed, then conditional is true. Non-material conditionals, like the examples given above, often make assertions about how the world would have been had things been somewhat different from the way that they actually are. The material conditional that merely makes an assertion about the way things as a matte of fact are does not logically necessitate anything about how thing would have been had the world been somewhat different. Unlike generalized material conditionals, however, laws do appear to support non-material “counterfactual” conditionals. Counterfactual conditionals are assertions about ways the world could have been, not assertions about the way it actually is. Since laws appear to support counterfactuals like the examples above about copper rods, a correct account of laws ought to assert more than just that a regularity holds as a matter of fact. To regard a claim as expressing a law is to take it as asserting a truth that would still be true had any of a broad range of contingent truths about particular states of affair been false. Another way to put this point is that statements of law go beyond merely describing actual regularities, they also tell us how things would be across a range of possible worlds. To sum this point up, we might say that laws have some “modal Umph” not enjoyed by the universally quantified material conditionals the Positivists associated with laws.
The Positivist’s C-Rules also often make use of non-material conditionals. For instance, the following might be given as a C-rule for having a charge of 1.5 volts:
C-rule for charge:
Conductor x has a charge of 1.5 volts iff, if conductor x were attached to a voltmeter, then the meter would register 1.5.
C-rules involving non-material conditionals are necessary for defining many theoretical terms. But they cannot be accommodated by the austere conditions for meaningfulness offered by the positivists.
Another way to get at the point is to note that non-material conditionals typically tell us what would be the case if certain non-actual conditions obtained. The conditions under which such claims are true cannot be fully analyzed in terms of the actual empirically knowable facts. But the positivist's criterion for meaningfulness requires that the truth or falsity of meaningful claims be analyzable in terms of truth functional operations on the empirically knowable facts.
Hempel’s deductive nomological account of explanation holds that explanations have the form of a deductively valid argument whose conclusion is the fact to be explained (the explanadum) and whose premises specify initial conditions and some law of nature linking those conditions to the explanadum. First, we can identify counterexamples in cases that fit the basic criteria of the DN account but do not explain. For instance:
1) Men who take birth control pills do not get pregnant.
2) Bruce is a man and he takes birth control pills.
3) Bruce is not pregnant.
Here the problem is that the facts offered as explanation are just not relevant to the explanadum. Other counterexamples reveal an asymmetry problem identified by Sylvain Bromberger:
1) Falling Barometer readings are followed by storms.
2) The barometer is falling.
3) A storm will follow.
We do think that an approaching storm explains a falling barometer reading. But not vice versa. In response to these problems of relevance and asymmetry, Hempel grants that his view recognizes many argument-like sets of claims as explanations that do not have any explanatory value to us. But this is a practical shortcoming we should not expect to overcome through philosophical analysis according to Hempel.